OCTOBER 27-28 MEETING OF THE CENSUS ADVISORY COMMITTEE
OCTOBER 27
Arnold Jackson, the Census Bureau’s “designated federal official,” started by announcing some Census staff changes, and thanking the committee members who agreed to serve as session discussants. Most of this meeting’s sessions started with a Census Bureau presentation, followed by discussant remarks, and then committee discussion.
Census Bureau Director Louis Kincannon explained that the 2010 Census Advisory Committee (CAC) is a new committee, intended to be more focused than its predecessor on the upcoming census. Following the 2010 census, he expects the committee might lie fallow for a period, but acknowledged that his successors might see a need for continuity. The new committee also has fewer members (down from about 40 to 25), and less duplication with other advisory committees. In selecting members, Kincannon noted that they focused on “data using” organizations. Another difference is that this committee reports to the Census Bureau and its Director, rather than to the Secretary of Commerce. Asked how often the CAC will meet, Kincannon said the objective is to meet as necessary—based on budget cycles, and the completion of census tests, and other relevant events. The CAC’s chair and vice chair positions are currently open, but the Bureau has candidates, and hopes to announce appointments soon.
As of the October 27-28 meeting, the 2006 budgets had not been passed, and Kincannon noted that the Census Bureau was operating under a continuing resolution that was sufficient to keep them moving along, but which would become problematic if extended beyond the November 18 expiration. The basic situation had been reported for some time. The House of Representatives approved full funding for the Census Bureau, but the Senate mark called for a major cut. If the Senate number held, Kincannon confirmed that the American Community Survey would be canceled for the rest of the decade, and they would prepare for a 2010 census with a traditional long form. Kincannon observed that the House was resolved to stand by its number, but noted the difficulty of the budget climate. About a week after the CAC meeting, the Census Bureau budget was passed at an amount close to the House mark—thus allowing the continuation of the American Community Survey.
Kincannon described Hurricane Katrina was a big event that has changed ways of thinking about a lot of things in Washington. At the Census Bureau, it has focused attention on the need to provide data relevant to events of this type, and in response, the Bureau has provided data describing the population and businesses in the impact areas prior to the hurricane. Already more proactive in such matters, the Bureau provided data of this type to FEMA in advance of Hurricane Wilma.
Kincannon described the strategic planning function now required of federal agencies, and that he has asked Nancy Gordon to take the lead position in this area (Associate Director for Strategic Planning and Innovation). And in a move called for by many at the last DCAC meeting, the Census Bureau has created the position of Chief Privacy Officer. Gerald Gates has taken the new position.
Kincannon concluded by noting the recent passing of Ed Goldfield, and describing his long career at the Census Bureau and the Committee on National Statistics.
Asked if the Census Bureau will measure the populations displaced by the recent hurricanes, Kincannon described how CPS respondents will be instructed to include information on hurricane evacuees staying in the household, and that the ACS question on residence one year ago is expected to provide additional information. In separate conversations, Census officials described ideas for estimating the impact on population sizes, but noted that the required data resources are not yet available. Your APDU representative can attest that the private sector data suppliers are in much the same situation.
Privacy/Confidentiality Update.
Gerald Gates commented that the position of Chief Privacy Officer is new to the Census Bureau, its functions are not. The Chief Privacy Officer advocates for privacy concerns, and communicates both internally and externally on privacy and confidentiality matters. Gates also will develop a preparedness plan to guide the Bureau’s response to privacy incidents.
Gates described a Privacy and Confidentiality Workshop that the Census Bureau conducted in March. The workshop included Census and external participants, who were led through an intensive adversarial exercise (led by former Census Director Vincent Barabba), designed to identify key issues to be addressed. In the future, Gates expects to report on and seek CAC input on emerging privacy/confidentiality issues, and the data Census Bureau’s data stewardship efforts.
CAC members sought to clarify the Chief Privacy Officer’s role—asking how the position will work with the Disclosure Review Board (DRB), and whether it will serve as a point of first contact for users seeking to negotiate data releases. Gates explained that he will interact with the DRB, but not be involved in the review of specific data products. Some surprise was expressed that the Chief Privacy Officer would not be more involved as a point of contact for data users, and would instead be focused on issues and processes.
Strategic Planning and Innovation
Nancy Gordon. Associate Director for Strategic Planning and Innovation
As described above, Nancy Gordon is the Census Bureau’s first Associate Director for Strategic Planning and Innovation. Gordon spoke briefly about the newly created position—describing it as full of opportunities—including, but not limited to the development of new data products. She referred the committee to her paper which describes the objective of ensuring that the Census Bureau is prepared to meet emerging data needs with relevant, innovative, yet fiscally responsible data products. Two examples are the need for data products measuring the impact of military base closings, and products describing daytime population. Daytime population (the subject of a recent Census Bureau press release) is a good example of how a new product, addressing different needs, can be developed without additional data collection.
Joan Naymark (U.S. Chamber of Commerce), the session’s discussant, commended the objective of leveraging census data, and the potential contributions to problem solving. And she stressed that the emphasis should be on problem solving rather than data use. Naymark also reported that her informal poll suggests that such efforts (and data products) will be favorably received. However, she cautioned that the Bureau needs to manage expectations and resources, and take care in the timing of its product roll-outs.
2010 Decennial Census Program Update
Preston Jay Waite. Associate Director for Decennial Census
Waite started with his own definition of the continuing (budget) resolution as the lowest amount anyone thinks you might get. He noted that the continuing resolution does not impact MAF/TIGER work, but that the ACS was moving ahead only cautiously, as the program requires significantly more funding in 2006 than 2005. Funding higher than 2005 levels is required to cover a full year of full implementation (versus 9 months in 2005), the inclusion of group quarters, and a methods panel to test alternative question wording. Waite described other 2010 activities hanging on the budget resolution, including the 2006 census test, and two large IT contracts—one on data capture, and another on the use of hand-held devices.
Among the findings of the 2004 census test are that hand-held devices seem to work well for automated re-interview, and enumerators like using them. The coverage improvement questions yielded mixed results—doing a good job of identifying erroneous enumerations, but not finding many more missed persons. The 2005 mail out/mail back test of content, question wording, and residence rules is on-going, and seeks to identify which options yield the highest response rates. The 2006 test involves alternative race and ethnicity questions and coverage measurement, but again, will proceed only if the budget permits.
One CAC rep wondered if the budget uncertainty puts the Census Bureau in a position similar to that leading up to the 2000 census, when they had to plan for two census methods—not knowing which would be permitted or mandated. Both Waite and Kincannon argued that they will not proceed with two paths. The current situation certainly differs in that there is no political battle over which method to use, but the reality is that the Census Bureau was describing alternative scenarios for the 2010 census – one with an ACS, and one with no ACS. Asked if, in the event of a canceled ACS, the 2010 census would have a long form, Waite said that is the plan, but that it would be costly.
Your APDU rep was unable to get the question on the record, but in a side conversation, asked Waite if a lack of funds for adding group quarters to the ACS would be enough to cancel the ACS. Waite expressed the view that an ACS without group quarters is not an option, as it would fall short of the goal of long form replacement. He also noted that if ACS funding lacked only the amount needed for group quarters, other options might be found. But he took the opportunity to expand on the delicate budget situation—noting that congressional appropriators might respond to a canceled ACS with further cuts in the census budget that could make it difficult to fund a long form in the 2010 census.
ACS Ramp Up and ACS Data Collection in Gulf Coast Areas
Marvin Raines. Associate Director for Field Operations
Marvin Raines started with an overview of the Field organization, noting that 95 percent of the Field staff is based away from the Suitland headquarters—scattered across the US in the regional offices, and the Jeffersonville processing center. He summarized their work with the “three verbs”—mail, call and knock—and quantified the enormous amounts of mailing, calling, and knocking (on doors) that they do in the data collection process.
Raines then described the field process for the ACS—including mail out, non-response follow up (both telephone and personal), post collection check-in, and processing. The ramp up to full implementation has been significant—requiring the addition of 400 staff for telephone interviewing, and 2,500 for personal visit interviews.
Turning to post-hurricane activities, Raines noted that they have been watching weather forecasts more closely than before, and are becoming more familiar with the activities of the US Postal Service, telephone companies, and other organizations. However, the top priority is ensuring the safety of Field personnel, and both Raines and Kincannon were pleased to report that while some Field staff are still displaced by the hurricanes, all are accounted for and safe. As for data collection processes in the impacted areas, Raines noted that they will definitely make adjustments to their normal processes, and he expressed confidence that they are doing this.
K.V. Rao.
Representative of the Race and Ethnic Advisory Committees
K.V. Rao drew a contrast between the CAC and the Race and Ethnic Advisory Committee (REAC) on which he serves. As Rao described it, the CAC role is to focus on the census process and its resulting products, while the REAC focus is on the persons who respond to the census—how they feel about the census and its questions. They are also concerned with quality of response and confidentiality issues.
Rao also expressed concern that the newly created Chief Privacy Officer position has no control over what data are released. Another REAC concern is with the possible inclusion of an ancestry question on the short form in 2010—that it could generate a lot of non-response and bad responses. Other concerns include the diversity of Census Bureau staff (including hiring, retention and promotion), the awarding of Census contracts to minority subcontractors, and a number of issues related to language.
Asked to elaborate on the concern over a short form ancestry question, Rao described concerns that the data could be misused—again, the perspective being that of the respondent. Another rep wondered if short form ancestry data (requiring processing from write-in responses) could be processed in time for the redistricting files. Jay Waite explained that the idea is not to report ancestry tables on the redistricting files, but to use the responses to improve the imputation of race, where race responses are missing.
MAF/TIGER Update
Robert LaMacchia. Chief, Geography Division
Bob LaMacchia started with background on MAF and TIGER – describing how the Master Address File works with the U.S. Postal Service Delivery Sequence File, and the Census Bureau’s Boundary and Annex Survey. He explained that Geography Division is currently reviewing the criteria for defining geographic areas for the 2010 census, and pursuing the Local Update of Census Addresses (LUCA) program, where a number of improvements are being tested. LaMacchia also described a Web TIGER program they are pursuing as an alternative to the media in which TIGER files have been disseminated in the past.
Asked if a post census local review is planned for the 2010 census, LaMacchia downplayed its importance, suggesting that the LUCA approach is more effective. When a rep suggested that some of the 2000 group quarters geocoding errors could have been caught with post-census local review, Jay Waite commented that such errors could have been better caught in the LUCA program. It was then observed that local areas that “slept through” LUCA would likely sleep through post-census local review as well. Clearly, some locals continue to be frustrated with the Census Bureau’s address verification efforts, and the Census Bureau continues to be frustrated with the participation among some local areas. When asked for an update on the suggestion for intermediate geographies (such as tract groupings) to enable the publication of more ACS data, Both LaMacchia and Waite noted that this is not a high priority, and there is not much happening on it.
Overview of Selected ACS Population Profiles
Douglas Hillmer. American Community Survey Office
The proposed ACS “Selected Population Profiles” product would provide a series of tables describing the characteristics of specific race and ethnic groups (such as Alaska Natives) and other population subgroups (such as persons age 60 and over), and Committee representatives were provided with example tables for review. Hilmer stressed that these data would not be derived from the detail file, but calculated from the microdata sample. Still, there is concern with disclosure avoidance, and subgroups must have a minimum population of 250,000 to have profiles published. Also, DRB restrictions requiring a minimum of 50 unweighted cases for each table would be applied – although this restriction would not apply to profiles based on 5-year average ACS data (see Note below).
Hillmer’s paper posed two questions: 1) whether the profiles should include estimates of total population, and 2) whether there are populations (not already iterated in SF4) that should be considered for profiles. Discussant Corinne Kirchner (American Federation for the Blind) recommended that population estimates be provided, although another rep observed that there would be three different totals for the one, three and five-year data. Kirchner also wondered if the disabled population could be presented as a subgroup rather than just a characteristic. And when she recommended tables with numbers, as opposed to percentages (as shown in the examples), Hillmer explained that the plan is to provide electronic tables allowing users to toggle between numbers and percentages.
Note: Although not covered in the CAC meeting, the Census Bureau is now applying restrictions to special tabulation products (now called custom tabulations), and plans some interesting restrictions for ACS data products. As a disclosure avoidance measure, data from custom tabulations are now rounded (usually to the nearest 5), and data are suppressed for areas with fewer than 50 unweighted records. No such limitations are described for ACS products, but plans call for the “collapsing and filtering” of ACS data based on tests for statistical reliability. If the data for a given table for a given area fail a reliability test, data could be provided for a collapsed version of the table. If the collapsed table fails the test, the data for that area would be suppressed. However, these restrictions would not apply to the 5-year average ACS data. The Census Bureau explains that the 5-year data are the core ACS product—designed to replace the decennial long form, which provides data without such restrictions. The 1-year and 3-year data are seen as more comparable to sample surveys, where reliability is a greater concern.
Coverage Improvement for the 2010 Census
Frank Vitrano.
Chief, Decennial Management Division
Frank Vitrano summarized some points from his paper on this topic. Among the many coverage improvement activities are efforts to assess the effect of communications on coverage (especially on Indian reservations), and improving the count of persons in households with more than six persons. The Census Bureau also is looking at the census residence rules—and instructions for explaining them on the form—as well as working on de-duplication methods, and automated address canvassing with hand-held devices.
Discussant Ilene Jacobs (California Rural Legal Assistance, Inc.) noted the persistence of differential undercount, and the “structurally ethnocentric” census operations. She stressed that partnership work with trusted local organizations needs to be underway now to reduce the number of missed whole units. She also wondered how many bi-lingual questionnaires the Census Bureau expects to need, and how to determine where they will be needed most. Jacobs cautioned that de-duplication efforts could inadvertently eliminate some units, and wondered how administrative records would be used to improve coverage.
Vitrano described the identification of large households as a potential application for administrative records. He then explained that in 2000, the Census Bureau intended to re-contact all households with more than six persons, but did not—apparently due to a lack of time. He confirmed that the plan for 2010 is to re-contact all such households. They are considering the advantages of personal visits to such households, but are concerned with the additional expense.
Content Testing
Claudette Bennett.
Population Division
Claudette Bennett described cognitive work on the race, ethnicity and ancestry questions in the 2005 Census Test. The work involved interviews with 212 (diverse) respondents, with objectives including the identification of ways to shorten the census questionnaire, and maximize self-identification of race and ethnicity. The test involved alternative forms of the race, ethnicity and ancestry questions to see if shortened race and Hispanic ethnicity question can be used in conjunction with a third question on ancestry. The interviews also tested reaction to alternative question wording, and the provision of examples—such as identifying Italian, Dominican, Chinese, Pakistani, etc. as possible ancestry responses.
The cognitive test results suggest that respondents do not need examples to help them answer the Hispanic or race questions, and were not necessary for understanding the ancestry question. Bennett noted that simple examples of ancestries were received more favorably than those that respondents considered “exotic.” The results also suggest the importance of a separate question on American Indian and Alaska Native tribes.
Discussant Helen Samhan (American-Arab Anti-Discrimination Committee) offered the opinion that the combination of the race, ethnicity and ancestry questions is a positive step. To address the perception issue (identified by Rao of REAC), she stressed the importance of making it clear that these are questions being asked of all Americans. She also suggested that the addition of ancestry to the short form could increase response to the census among some difficult-to-enumerate groups who otherwise could report their identity only on the long form—or now the ACS. At this point, Debbie Griffin (Census Bureau Demographic Statistical Methods Division) came forward from the audience to point out that ancestry responses from the ACS are actually more complete than those from the long form. Samhan acknowledged the problem that some groups have with the combined approach, but observed that Arab Americans might be especially suspicious of the ancestry question on a form not part of the decennial census. In responding to a question on the difficulty recent immigrants might have with a write-in response, Bennett noted that some questions left unanswered by the cognitive tests, will be addressed by the 2005 national test.
2006 ACS Content Testing
Wendy Hicks. Decennial Statistical Studies Division
The 2006 ACS Content Test is designed to determine ACS content for 2008—the first year that both 3-year and 5-year ACS data will overlap with the next decennial census.
Wendy Hicks explained that the test’s primary objectives are to 1) determine if changes in response categories and question wording improve the quality of data collected for specific content areas, 2) test whether changes in the layout of the mail form impact response at the unit or item level, and 3) determine if the ACS can contain mailing costs—without negatively impacting data quality—by dropping the questionnaire instructions booklet. Also to be tested are new content items related to recent or anticipated legislation, and a number of alternative questionnaire formats.
Discussant Arturo Vargas (National Association of Latino Elected and Appointed Officials Educational Fund) offered a number of observations on the paper, and stressed the importance of understanding the reasons for item non-response – such as the fatigue factor in answering long questions. With respect to the possibility of new ACS questions, Vargas asked what some of these might be. Hicks identified several, including health insurance, marital history, and the service connected disabilities of veterans.
Public Comment
There were no public comments.
Communications Planning for the 2010 Census
Edison Gore.
Assistant Division Chief for Planning and Outreach, Decennial Management
Division
Edison Gore described the success and innovations of the 2000 census communications program – including paid advertising, extensive partnerships, and innovations such as the Director’s response rate challenge to local areas. The Census Bureau is now embarking on the 2010 program, which they hope will be at least as effective as the 2000 program. Gore described the plan for an “Integrated Communications Program” based on research, testing, lessons from 2000, and input from industry experts. The number one goal is to improve response rates through education, creating awareness, and motivating participation. The Census Bureau also hopes the ACS will boost response by eliminating the drag of the long form’s reduced response rate – and by providing data that can be used to improve the targeting of census communications. Gore explained that the LUCA program is part of the communications effort, as the Bureau seeks to make local areas aware of the program, and to encourage participation.
Discussant Ed Spar (Council of Professional Associations on Federal Statistics) complimented the Census Bureau on the 2000 program, but wondered if there are specific measures of advertising effectiveness. If not, he recommended that such measures be built into the 2010 program. Spar also asked about the focus groups with faith-based organizations, and the reference to incentives.
Gore reported that they did evaluate the 2000 advertising campaign, and while the results confirmed success in increasing awareness, it is more difficult to measure the impact on motivation. Gore also explained that outreach to faith-based organizations is not new, as they are seen as community gatekeepers. And the reference to incentives relates to promotional items such as census balloons and posters, but also drop-in articles for use in local newspapers. Jefferson Taylor (Associate Director for Communications) responded to a question on the Census in the Schools program, noting that it might be difficult to fund, and that they are looking at the possibility of funding it through a foundation partnership. Taylor also noted (in response to a recommendation) that they are looking at the possibility of extending the Census in the Schools program to include adult education. At this point, Jay Waite observed that everyone likes partnerships, but they are tough to sell to the appropriators.
Paul Pisano.
Chief, Congressional Affairs Office.
John Cuaderes. Subcommittee
on Federalism and the Census.
David McMillen. Committee on Government Reform
Paul Pisano started the session with a review of the 2006 budget situation and numbers, and a summary of what was at risk if the Senate mark (the funding cut) had been adopted. He described the ongoing challenge of turnover among congressional members and staff devoted to census oversight, the constant need to educate, and the challenge of demonstrating the relevance and importance of census data. As several others had throughout the meeting, Pisano described efforts to showcase the relevance and importance of census data to hurricane response. He also noted that the recent press release on daytime population data had attracted some attention in Congress.
John Cuaderes (congressional staff from the majority side) was unable to attend, so David McMillen spoke next.
McMillen described 2005 and 2006 as years when critical decisions must be made for the 2010 census, and commented that it is scary that we’re sitting here not knowing what will happen. He described the budget showdown as a House vs. Senate thing, but also a Republican vs. Democrat thing. Normally this is the point where the politicians start to make deals, but McMillen cautioned that in the post-Katrina, post-Iraq environment, there is a new sense in Congress that “we really ought to pay for things.” The balanced budget voices are emboldened, budget cuts are in the works, but it is not yet clear what will be cut. The chaos on the Hill, he said, is at a level we haven’t seen in a long time, and it puts the census in danger. McMillen closed by announcing that this was his last day with the Committee on Government Reform, as he will be taking a new position at National Archives.
Asked about possibilities for his replacement, McMillen said it will be Michelle Asch, who is not new to the census. Asked about the possibility of losing both the ACS and the long form, McMillen said he did not expect the worst case scenario, but described the danger of limping along year to year with a marginally funded ACS until it is too late to go back to the long form. With this risk in mind, he commented that one could wonder if it might be better to terminate the ACS in 2006. Asked how such a worst case scenario might play out over the following decade, McMillen suggested that census data would be missed, but it would be too late, and we would all be worse off for it. Jay Waite commented that there might be renewed interest in the ACS, but that the costs of re-launching it would be far higher than what we are used to spending, and there would be reluctance to spend the money. Waite also expressed doubt about how effectively the Census Bureau could ramp up again, and contended there would be negative impacts on the 2020 census.
During the Q&A portion of his presentation, David McMillen received well wishes and expressions of thanks from many CAC representatives, and the Committee passed a motion approving a more formal expression of the committee’s gratitude.
There followed a proposal for a CAC letter in support of Census funding. There was doubt that CAC reps could get approval from their organizations in time for such a letter, but also concern that such a letter might violate the Committee’s charter. Follow up communication from the Census Bureau made it clear that such contact with congressional appropriators, in the name of the Census Advisory Committee, would be “out of scope” of the Committee’s “advisory” role.
Because so much is likely to happen before the spring CAC meeting, there was a question about establishing working groups. There were expressions of support from Census Bureau participants, and Arnold Jackson assured that the suggestion would be relayed to the Director.
Recalling how valuable site visits and participant experiences had been in earlier Decennial Census Advisory Committee activities, one rep inquired about the possibility of such experiences for the new CAC. Jay Waite agreed that this is a great idea. There is always the question of funding but he indicated that they could probably make this happen.